[Talk-ca] Canada building import - Simplification discussion

[Nate Wessel]

I'm changing the subject line to try and retain some clarity for the 
mailing list.

James, thanks for the stats! I'm surprised this didn't remove more points.

Steve, I mentioned earlier that 20cm is where I happened to draw the 
line for the data we imported in Hamilton County, Ohio and I'm not 
totally sure even that was ideal. A bit more than half of the buildings 
we've been importing there have one (damned) extra node which I've been 
trying to remove manually. I may have been a bit too conservative there, 
as I didn't want to lose any part of the geometry. Sometimes there are 
tiny bay windows and the like.

I'm wondering if the scope of this data warrants some analysis of how 
simplification (and perhaps data quality generally) varies 
geographically. It seems quite likely that every municipality would have 
it's own quirks, and we might want to treat some places differently.

Best,

Nate Wessel
Jack of all trades, Master of Geography, PhD candidate in Urban Planning
NateWessel.com <http://natewessel.com>

On 1/19/19 7:24 PM, OSM Volunteer stevea wrote:
>> On Jan 19, 2019, at 3:47 PM, James <james2432 at gmail.com> wrote:
>> Resending because these emails are getting over 40KB in size and talk list is spazzing out:
>> Original:
>> 9.263 Average points per feature
>> Points:20346517
>> Features:2196329
>>
>> Simplified (20cm): 8.425 Average points per feature
>> Points:18504036
>> Features:2196329
> Choosing 20cm or higher or lower is where the "improvement?" line gets drawn.
>
> In round numbers of nodes, call this a "10% reduction."  The question is, are the results "improvement?"  If so, it seems worth doing.  (Only one person's opinion, of course).
>
> Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20190119/4ec824fb/attachment.html>